The last U.S. jobs report included big revisions. Here’s why.
Job growth was weak in July, and revisions to May and June numbers suggest that it reached a near standstill over the summer.
The most important indicator of the economy’s health, hands down, is the job numbers released monthly by the U.S. Bureau of Labor Statistics. So, it was big news when the July job numbers unambiguously showed the economy is sputtering.
In response, President Donald Trump fired the commissioner of the Bureau of Labor Statistics, saying she had rigged the numbers for political purposes. By all accounts, there is no evidence of her doing so.
The president wasn’t pleased with the weak July job growth, but the downward revision to the job gains in May and June appears to have been particularly triggering. The data now suggest that job growth has reached a near standstill this summer, and recession risks are uncomfortably high.
The downward revisions were unquestionably large, but they are not unusual when the economy is in trouble, as it is now. In fact, the revisions provide helpful information regarding how close the economy is to the brink.
This is wonky, but let me explain. The payroll job numbers we are focused on are based on a business survey. More than 600,000 worksites across all industries and regions of the country that employ more than one-fourth of all workers participate in the survey, which is conducted during the week that includes the 12th of the month.
Because not all businesses respond quickly, their responses are picked up in the subsequent two months. Approximately 60% of businesses respond fast enough to be included in the first estimate, and more than 90% by the third estimate.
Most of the time, when the economy is moving more or less in a straight line, there isn’t a meaningful difference in the performance of businesses that respond early and those that respond later. However, when the economy turns down, businesses that respond later feel the downturn more, which is reflected in their weaker payrolls. Thus, the downward revisions.
The big downward revision to the job numbers in July thus demonstrates that the economy indeed has a problem.
Recent events raise good questions about the quality of economic data and what should be done to improve it. The data could undoubtedly be more accurate, timely, and comprehensive.
The most serious problem plaguing nearly all surveys is falling response rates, which have worsened since the pandemic. This may be related to survey fatigue. Think about how often you are asked to rate the product you purchased or the service you experienced. It’s gotten to be too much.
Privacy and cybersecurity are also concerns when responding to surveys. Privacy may be increasingly top of mind when responding to a government survey, particularly when there are reports of data collected by the government being misused. Increasingly sophisticated cybercrime, potentially empowered by artificial intelligence, may also have a chilling effect on response rates.
Ironically, the response rates to the survey of businesses have been OK. However, responses to other government surveys used to construct critical economic data haven’t held up nearly as well.
Consider the all-important Consumer Price Index (CPI). Due to budget and staffing cuts in recent months, there aren’t enough BLS surveyors to canvas prices. Over one-third of the prices of goods and services included in the CPI are now imputed — not measured directly, but estimated based on the prices for products that have been canvassed. Before these cuts, only about one-tenth of the prices in the CPI were imputed.
This provides a good example of the costs of less accurate economic data.
When global investors purchase U.S. Treasury or other bonds, they assess future inflation. If they expect inflation to be higher, they want to be compensated with a higher interest rate. Same if they are more uncertain about future inflation, which would be so if the CPI is no longer as accurate. Even a tiny increase in interest rates will cost us big time as taxpayers and consumers, as we have tens of trillions of dollars of bonds outstanding.
A different solution to the response rate problem has been floated, at least for the job numbers. Namely, wait a few months until more responses come in. For example, instead of the BLS releasing an estimate of August job growth in early September, it would wait until November to collect more responses before releasing an estimate for August.
Consider what this would mean for the Federal Reserve Board when deciding on interest rate policy. As it is now, if the economy slips into recession, the Fed would immediately cut rates to forestall the downturn or cushion the economic blow. But if officials had to wait until November to get the August job numbers, they would likely be much slower to respond to the faltering economy, with potentially serious consequences.
Not surprisingly, economists, like me, want better economic data. The point is: You should, too. The nation’s economic well-being and your financial health depend on it.